Accurate sound measurement requires technical equipment that must be calibrated and used correctly.1 Unfortunately, the unit used most often in measuring and describing sound levels — the decibel (dB) — is complicated and confusing for several reasons. Chief among these is that decibels are a logarithmic scale rather than a linear one, and are often also “weighted” to highlight certain frequencies (see below).
For these and other reasons, many noise-reduction advocates recommend using perceived volume levels — i.e., how sound is actually experienced by those (including enforcement officials) exposed to it — as the basis for noise-related public policies, rather than a technical metric such as decibels. And in fact, Providence’s sound ordinance includes such a standard, which is “audible at 200 feet from the source.”
Logarithmic scale
As noted above, decibels are measured on a logarithmic rather than linear scale. A linear scale increases at a constant rate, such that if the numerical value “X” is twice as high as value Y on a linear scale, it is two times more intense. Graphs of linear measurements are straight lines that rise or fall at a single rate, making the relative intensity levels of different sources easy to compare.
By contrast, a logarithmic scale is exponential, which means as the numerical value increases, the rate of change also increases — i.e., the intensity level rises at an increasing rate. For this reason, graphs of logarithmic functions are curved lines that get steeper (more vertical) as they increase or decrease, rather than a straight line from one value to the next.
For sound, this means that a decibel level that’s twice as high as another one isn’t double the volume, but multiple times louder. As a result, relatively small changes in dB values are actually quite substantial. Increasing the volume of a sound from 50 dB to 60 dB isn’t a 20% rise, but actually three times louder.
Weighting
To further complicate matters, most sound-level measurements also use what are known as “weighted” decibels that de-emphasize specific parts of the sound spectrum, rather than registering the entire acoustic range the same way. These include A-weighted decibels, less common C-weighting, and rarely used Z-weighting (which is actually an absence of weighting — i.e., “raw” or unweighted data).
Most modern noise-related public policies such as sound-level regulations and enforcement measures use the “A-weighted” or dB(A) scale — the blue line on the graph below — which focuses on the middle and upper frequencies prevalent in the human voice (which is different from the human hearing range), and was originally devised for occupational settings such as factories and other industrial workplaces.
Our research indicates that A-weighting derived from the idea that if exposure to excessive noise in a workplace damaged an employee’s hearing to the point that it was difficult for them to understand other people’s voices (e.g., their co-workers, supervisors, etc), they would arguably be less employable and so could sue the company seeking monetary compensation for future lost wages. Thus, for reasons of legal liability, it was important to measure noise in the range of human speech.
As a result, A-weighting deliberately excludes much of the low-frequency “bass” that comprises a disproportionate amount of urban noise sources — mufflers, subwoofers, leafblowers, and other vibration-heavy sounds — that travel farther and more easily penetrate buildings than higher-frequency sounds. This means that A-weighting is actually uniquely ill-suited for measuring urban noise — even as it remains the predominant sound metric in U.S. municipal ordinances and other regulatory frameworks.
As the graph above indicates, C-weighting is far more accurate in registering the low-frequency sounds that A-weighting omits, and closer to Z-weighting (i.e., no weighting).2 For that reason, the Noise Project uses C-weighting to depict sound levels that Providence residents actually experience on a daily basis.
_________
1 Cellphone applications can indicate general sound-level ranges (e.g., “above X decibels”), but are not sufficiently accurate to rely on as an absolute value (“the noise was X dB”) or to use as evidence in legal proceedings.
2 For this reason, C-weighting is also used to calibrate professional sound-level meters to ensure that the level registered by the meter matches a test tone generated as a reference value.